epidemiological studies of leukemia in persons …...epidemiological studies of leukemia in persons...

Epidemiological Studies of Leukemia in Persons Exposedto Ionizing Radiation*

L. H. HEMPELMANN

(Division of Experimental Radiology, University of Rochester School of Medicine and Dentistry, Rochester, N. Y.)

SUMMARY

After reviewing the epidemiological studies of persons exposed to ionizing radiation,the author concludes that there is no question that radiation exposure in man is associated with an increased incidence of leukemia and that a cause-and-effect relationshipmust be accepted. Such an association holds in the higher dose range regardless of (a)sex and age, (6) the amount of body exposed (provided a substantial amount of blood-forming tissue is irradiated), and (c) whether or not the radiation was administered as asingle dose or intermittently over a long period of time. In two studies which have anadequate sample size and a sufficient number of cases in each of several exposure groupsto permit analysis, there is a proportionality between leukemia incidence and dosage inthe middle dose range. Probably because of the small number of cases studied, such arelationship was not observed in children treated for thymic enlargement in infancy.

The author believes that the available data are insufficient to permit conclusions tobe drawn concerning the relationship between leukemia incidence and exposure to lowdoses of radiation. Thus, in the British series of spondylitics, only one case of leukemia(excluding a second case with extraspinal radiation) occurred in the relatively few patients receiving less than 500 roentgens to the spinal marrow. In the Japanese, uncertainties regarding actual dose values as well as the small number of cases in the exposure zones between 1,600 and 2,000 meters preclude extrapolation of the curve ofdose-versus-incidence to the low dose range. The other studies reviewed in this paperdo not lend themselves to analysis in the low dose range either because of the smallnumber of cases or dose uncertainties or, in the case of prenatal exposures, because therole of radiation in the development of the disease has not been established.

Although there is some suggestion that children are more sensitive to the leukemo-genic effects of radiation than adults, the author concludes that the present retrospective studies of prenatal exposure of children dying of leukemia or cancer do not establish exposure to diagnostic doses of radiation as an etiologic factor.

The difficulties of carrying out epidemiological studies in man to show an increase ina rare event such as leukemia have been pointed out. The size of the group needed forstudy, the long latent period between exposure and study and the choice of proper control groups present serious problems, especially in the low dose range. In view of ourneed for information about radiation hazards in the low dose range, it is emphasizedthat restraint must be exercised to avoid reading more significance into the availabledata than is justifiable. This does not mean that the possibility of radiation hazardsexisting at low doses should not be recognized. Clearly, unnecessary exposure of largegroups of people should be reduced to a minimum.

* Given as part of a symposium on Radiation in part, by a grant from the United States AtomicCarcinogenesis, American Association for Cancer Energy Commission.Research, Atlantic City, April 10, 1959. Supported, Received for publication August 24, 1959.

18

Research. on January 27, 2020. © 1960 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

HEMPELMANNâ€”IonizingRadiation and Leukemia 19

An accurate evaluation of the risk in man's ex

posure to small doses of ionizing radiation can bebased only on the study of large groups of personsknown to have had such exposures. The vastamount of data from experiments with irradiatedanimals and with simpler biological systems hasprovided important information concerning themanifestations of radiation injury and the mechanisms by which such injury occurs; as has beenpointed out repeatedly, most recently by Mole(21), such information cannot be extrapolated directly to man, at least insofar as the quantitativeresponse to given doses of radiation is concerned.Because of our great need for information aboutthe hazards of exposing man to the increasing andwidespread levels of radiation throughout theworld, it is important to consider carefully thedata now available from studies of irradiated human beings to assess the state of knowledge in thisfield. It is hoped that careful analysis can help toclarify some of the confusing and conflicting opinions that now appear in the medical literature andlay press. Similar critiques have been publishedpreviously by the United Nations Scientific Committee on The Effects of Atomic Radiation (33),Brues (2), Kaplan (11), and Loutit (14).

Before considering the epidemiological studiesof groups of people exposed to radiation, it is wellto point out the limitations involved in this typeof investigation. Three serious limitations shouldbe mentioned. The first is concerned with the sizeof the sample needed for study. Obviously, thenumber of subjects needed depends upon the magnitude of the effect to be detected. If one wishes todemonstrate an increase in a rare, spontaneouslyoccurring event, such as the development of leukemia, or to show a possible shortening of life bya small fraction of the total life span, it is obviousthat large numbers of people with known exposures must be surveyed. Invariably, such largegroups of people, usually numbering thousands,present serious problems for study. The secondlimitation depends upon the long interval betweenexposure and time of study. The procedure usuallyemployed in these studies is to survey a group ofpeople exposed to radiation many years before.Because of the elapsed time, a certain fraction ofthe group inevitably will be difficult or impossibleto locate. Also, certain of the persons contactedmay provide medical information which will bedifficult to confirm at the time of the study, andothers will have had additional uncontrolled exposure to radiation.

The final limitation is concerned with the planning of the epidemiological studies. Becausegroups of exposed persons satisfactory for study

are uncommon, the investigator has no choice butto accept the experimental design offered by thesituation. This is usually far from ideal in manyrespects. For example, the exposure data may beinadequately documented or impossible to calculate accurately. Also, the exposed individualsusually constitute a selected group for which thereis no available control group comparable in all respects except for radiation exposure. In this case,the selection of appropriate controls will be a matter of judgment, and this will be greatly influenced by practical considerations. The use of morethan one group to control different characteristicsof the exposed sample is often desirable, but thisadds to the time and expense of an already expensive and tedious study. Thus, it is apparent that,in the case of epidemiological studies of irradiatedpersons, it is impossible to reach the precise conclusions which can be deduced from well plannedanimal experiments.

With these limitations in mind, the author willnow consider data from the five groups of irradiated individuals most intensively studied from thepoint of view of leukemia incidence and radiationdosage. The studies in children will be discussed indetail, because they contain new data only recently available.

EPIDEMIOLOGICAL STUDIESCHILDRENRECEIVINGPARTIAL-BODY

EXPOSURETOX-RAYSInfants treated with x-rays for thymic enlarge

ment.â€”In 1955, Simpson, Hempelmann, and

Fuller (30) published the preliminary results of astudy of the incidence of neoplasia in a group ofchildren mainly from Rochester, New York. Thisseries of 1,722 children had been treated in infancywith x-rays for enlargement of the thymus glandduring the period 1926-1951. In their last report,

Simpson and Hempelmann (29) showed that manymore cases of malignant disease including leukemia occurred in the treated group than would havebeen expected on the basis of the number of children studied and the age-specific cancer incidencerates in upstate New York. Recently, Simpson(28) has studied 671 additional children treatedfor thymic enlargement in Buffalo, New York. Inthe 87 per cent of children traced from the composite group of 2,393,21 cases of cancer were foundinstead of the 3.6 cases expected to occur in thetotal group. Nine and possibly ten (one case unconfirmed) children had died of leukemia, whereasone death was expected. Thyroid carcinoma accounted for most of the other excess cases of cancer. There was not a significant difference betweenthe expected and observed number of cases of can-



20 Cancer Research Vol. 20, January, 1960

cer or leukemia in 2,722 untreated siblings of thestudy children. When the cases were divided according to the method of treatment, the childrenwith leukemia seemed to be randomly distributedamong the subgroups. More than half of the casesof other types of cancer occurred in one subgroupof 268 children. These children had been treatedwith 110-150 roentgens of 140 kvP x-rays througha 9-cm. circular port to the front as well as to theback of the chest. The treatment was repeatedonce and sometimes twice at biweekly intervals.The radiation factors used in treating the childrenin the other subgroups varied considerably duringthe 25-year period during which treatments weregiven, but, in general, the doses and port size became smaller in the later years. The smallest doseand port size used in any subgroup was 50-75 rthrough a 6 X 8 cm. port, usually to the anteriorchest.

The doses in roentgens as measured in air wereknown or were calculated from the radiation factors for all but 299 children. Four of the knowncases of leukemia occurred in 1,050 children receiving a cumulative dose of less than 200 r, whereas five cases were found in 1,025 children givenmore than 200 and usually less than 600 r. In contrast, all cases of other malignant neoplasia occurred in children receiving 200 r or more. Theaverage time between exposure and death fromleukemia was 5.3 years.

In addition to the study just reported, othersimilar prospective surveys of children treatedwith x-rays for thymic enlargement are now beingcarried out in this country. In a preliminary reportof a survey of 1,131 children so treated in a Bostonhospital (31), Snegireff states that he observed nocases of leukemia in 148 treated children or in 185untreated controls. In discussing Simpson's paper(28), Maloney mentions a follow-up of 125 of 700children treated in Boston. He did not find anycases of leukemia in this group but observed sevencases of thyroid neoplasia including two malignancies. Retrospective studies of a questionnaire typedesigned to obtain, among other information, thehistory of prior x-ray exposure of children withleukemia showed that only a very small percentagehad been treated for thymic enlargement. InGreat Britain, where the practice of treating thymus glands was abandoned many years ago, Stewart and her colleagues (32) obtained a history oftherapeutic irradiation (none for thymic enlargement) in only five of 677 children dying of leukemia. Snegireff (31) quotes Farber as having obtained a history of thymic irradiation in only five of856 children with leukemia seen at the TumorClinic at the Children's Cancer Center in Boston.

Polhemus and Koch (25) report a history of x-raytreatment to the thymus in 4.3 per cent of 261 children with leukemia seen at the Los Angeles Children's Hospital, in comparison with 0.8 per cent of

children so treated in a comparable group of controls without malignant disease. Murray and colleagues (22) reported that eight of 65 childrendying of leukemia in Rochester, New York, had ahistory of x-ray treatment (five for thymic enlargement), whereas none of 65 matched dead controls and only two of 175 living siblings had sucha history.

In evaluating these observations, it seems clearthat further studies of a prospective nature mustbe undertaken to establish the true incidence ofleukemia in children given thymic irradiation, particularly as far as the relation of the incidence todose and port size is concerned. In the survey inupstate New York, the number of leukemiadeaths, although not large in absolute terms, isunquestionably greater than that in their untreated siblings or in children of the general population.Since the state of the thymus gland is unknown forthe sibling group and is, for the most part, normalby definition in children of the general population,it is clear that this study does not differentiate between the association of leukemia and (a) priorexposure to x-rays or (6) the medical conditiondiagnosed as thymic enlargement. Because it isimpractical, if not impossible, to obtain an idealgroup, i.e., a series of children diagnosed as havingthymic enlargement at birth not treated withx-rays, it has been necessary to initiate studies ofchildren irradiated for other medical conditions inan effort to distinguish between the associatedand, possibly, leukemogenic factors.

Infants with normal-sized ihymus glands givenx-ray treatments.â€”Agroup of 1,564 children treated with x-rays in Pittsburgh was studied in 1948by Conti and his colleagues (3). Ninety-six percent of these children were shown to have thymusglands of normal size at birth. The radiation factors were uniform in the entire group, being: 140kvP x-rays; 3 mm. Al filtration; STD, 8-9 cm.;port size, 4X4 cm.; dose increments of 50-75 rat 1- to 2-day intervals. Eighty-eight per cent ofthe children received 75-300 r (usually 150 r) tothe region of the manubrium; the remainder received 200-450 r.

Ninety per cent of the children were contactedagain 11-18 years after therapy by Conti and hiscolleagues (4). The names of those not locatedwere searched for carefully in the death records ofthe Pennsylvania State Department of Health.Four cases of malignant disease were expected tooccur in this group, but not one was found; the one



HEMPELMANNâ€”Ionizing Radiation and Leukemia 21

expected case of leukemia was not observed. Therewas not a significant difference between the number of expected and observed cases of cancer andleukemia in the untreated siblings.

The failure to observe any cases of neoplasiawhen only four (including one leukemia) were expected is not significant, particularly since one-tenth of the group was not located. In view of thenegative observations, one can only conclude thatthere is no evidence for an increase (a doubling,perhaps) in the cancer rate in the treated childrenor a greatly enhanced leukemia frequency, i.e.,5-10 times that in the general population. Such

increases probably would have been detected inthis study. It is interesting, however, to comparethe findings in the traced children of this serieswith those reported in the New York study. Sucha comparison can be made directly without beingconcerned about the 10-13 per cent untraced chil

dren in each group. The difference in the numberof cases of leukemia in children receiving less than200 r (none in 1,375 children of the Pittsburghseries and four in 1,050 children of the New Yorkseries) is significant at less than the 5 per centlevel. One cannot say whether the absence of excess cases of leukemia in the Pittsburgh childrenis due to (a) their normal-sized thymus glandswhich, conceivably, did not predispose them todevelop leukemia, or (6) the smaller port size usedin treatment. Although the first possibility cannotbe excluded, it is entirely possible that the smallport size could be responsible for the difference infindings. A simple calculation shows that the volume of tissue irradiated in the case of the Pittsburgh children was one-third that of the NewYork State children treated with the smallestports. The exposed tissues include the blood-forming cells of the sternum and anterior portions ofthe ribs as well as those of the thymus gland andmediastinal lymph nodes.

Children given x-ray treatments for benign conditions.â€”To avoid complications introduced by considering only children given x-ray therapy to themediastinal region, a study was made of 6,473 children in Rochester treated with x-rays for variousbenign conditions (22). This group of children isbelieved to represent the vast majority of childrenso treated in Monroe County in the past 25 years.The only cases known to have been missed are thepatients of one radiologist who destroyed his therapy records upon retiring in 1944. The number ofcases of leukemia in this group was determined bychecking these names against the cancer recordsin the Division of Vital Statistics of the New YorkState Department of Health. In addition, thefamilies of all children dying of leukemia in this

county since 1940 were interviewed to obtain ahistory of previous radiation exposure.

Considerable bias was introduced deliberatelyinto the study to raise the number of expectedcases of leukemia and to lower the observed number. Nevertheless, the difference between the eightobserved and two expected cases is significant(P < 0.001). Five leukemia deaths occurred in2,750 children treated for thymic enlargement, twooccurred in 75 children treated for pertussis, andone in 1,073 children given x-ray treatments to thehead and neck region, mainly for lymphoid hyper-plasia of the nasopharynx. No deaths of leukemiawere found in 2,460 children treated with superficial x-rays for benign skin lesions. The fact thatthe number of observed cases exceed the expectednumber in three categories representing unrelatedmedical conditions including two cases in 75 children treated with x-rays for pertussis would seemto suggest that the high voltage x-rays used intreatment were the etiologic factor in the subsequent development of leukemia.

Discussion.â€”In evaluating the data obtained in

these three studies, it seems clear that an association has been established between radiation exposure and subsequent leukemia in certain childrentreated with x-rays for thymic enlargement butnot in the case of infants with normal-sized glandsgiven small doses of x-rays through a small port.

The occurrence of two cases of leukemia in 75 children treated with x-rays for pertussis and o ne casein about 1000 children given x-ray treatments to thehead is not significant in itself but fits the patternof an increased incidence of leukemia in irradiatedchildren. The occurrence of only twelve cases inapproximately 9,000 children studied emphasizesthe scope of the studies needed to establish a possible quantitative relationship between exposureand leukemia incidence, especially since port size,as well as dose, may be an important factor. In thenine cases observed in children treated for thymicenlargement, there was no obvious correlation between dose and incidence.

It should be emphasized that the doses of x-raysassociated with the considerable increase in leukemia frequency in these children is far less than thethousands of roentgens formerly believed to benecessary to induce malignant change.

PATIENTS WITH ANKYLOSINGSPONDYLITISGIVENX-RAY TREATMENTSTO THE SPINE

Court-Brown and Doll (5) have published athorough study of the cases of leukemia andaplastic anemia in 13,200 patients treated withx-rays for ankylosing spondylitis. They reviewedthe death certificates and studied the clinical and




pathological data of all patients suspected of dyingof leukemia or aplastic anemia. In addition, theyobtained dosage records from which they calculated the mean dose to the spinal marrow and thewhole-body integral dose of a large sample of thecases. They concluded that 32 proved and fiveprobable cases of leukemia occurred in this groupof patients as well as four cases of aplastic anemia.When the cases are divided according to the meandose to the spinal marrow or to the integral dose,there is a definite relationship between dose andleukemia incidence. The exact shape of the curveof incidence vs. dosage plotted on an arithmeticscale depends upon whether the mean spinal marrow dose or the integral dose is used and uponwhether or not the cases receiving extraspinal radiation are excluded. The curve is clearly linear inthe middle three dose groups expressed either interms of mean spinal marrow dose or integral dose.In the higher dose range, the curve departs sharplyfrom linearity unless the incidence in all cases receiving more than 1,250 roentgens is averaged as asingle point. There has been considerable discussion of how the curve should be drawn below 500 ror 7.5 megagram r, particularly with regard towhether or not there is a threshold dose belowwhich there is no increase in the leukemia incidence. The most favorable plot for the interpretation of linearity is found when the leukemia incidence of 35 cases of leukemia, excluding two coexistent cases, is plotted against mean spinal marrow dose. It should be emphasized that the onlypoint below 500 roentgens is based on two cases oflymphocytic leukemia. One of these patients developed chronic lymphocytic leukemia after receiving a mean marrow dose of 471 r; the other developed lymphocytic leukemia after a series oftreatments which delivered 113 r to the spine. Thetotal integral dose including that given extra-spinally to the second patient exceeded that givento the first patient, i.e., 12.2 vs. 18.3 megagram r.

It is interesting to note that, of the 50 cases ofleukemia in patients treated with x-rays for spon-dylitis, including those in Court-Brown and Doll's

series, 38 have been examples of the acute disease(7). Only one of the eight cases of chronic leukemia, presumably the patient mentioned above, hasbeen lymphocytic in type.

This report is clearly one of the most valuablepublished to date, with the dosage factors beingparticularly useful. The linearity of the curve inthe middle dosage range is unquestionable, as isthe curvilinearity in the higher dose region undermost conditions of plotting. However, there hasbeen considerable criticism of the attempts to extrapolate the data to the low dose region. This in

cludes the author's own mention of the dissimilari

ty of the curves obtained when the leukemia incidence is plotted against the integral dose or againstthe marrow dose. Also criticized are the smallnumber of cases below 500 r (2), the use of thespontaneous leukemia incidence in general population as a point on the curve at dose zero (24), andthe unknown role of age-susceptibility in the development of leukemia (19). In connection withthe attempted extrapolation of the incidence datato low doses, a word should be said about the non-irradiated controls available for comparison withthe study group. Since there appears to be a stronghereditary factor in ahkylosing spondylitis (23),the leukemia rate in the population-at-large cannot be accepted as the rate for the untreated patients. This is illustrated by the report of Abbattand Lea1 showing an association between untreated rheumatism and leukemia. The only availablecontrol group of 399 spondylitic patients not treated with x-rays is too small to be of use in determining the leukemia rate, since the absence of anycases when less than 0.2 case is expected is notmeaningful.

In view of the limited data in the lower doserange and the lack of an appropriate controlgroup, it seems reasonable to conclude that thisstudy does not provide evidence (a) to determinethe true leukemia incidence in the dose range below 500 roentgens or (6) to ascertain whether ornot a threshold dose for leukemia induction exists.It is interesting to note that almost all cases occurred in patients receiving more than 500 roentgens to the spinal marrow in contrast to the findings in leukemic children, almost all of whom received less than this dose when measured in air.One cannot determine whether this difference isdue to the small number of patients receiving lower doses (about 3,000 receiving less than 500 r),age-susceptibility, treatment technics, i.e., protracted vs. one or two treatments, or to the volume of blood-forming tissues exposed.

JAPANESEPOPULATIONSRECEIVINGTOTAL-BODYRADIATIONFROMATOMICBOMBEXPLOSIONSA number of reports published since 1951 by the

Atomic Bomb Casualty Commission have dealtwith the increased incidence of leukemia in theJapanese survivors of the atomic bomb explosionsin 1945. The results of these studies have beensummarized by Heyssel and his colleagues.2 That

1J. D. Abbatt and A. .1.Lea, Leukemogens, Lancet, 2:880,1958.

1R. Heyssel, A. B. Brill, I,. Woodbury, E. T. Xishimura,T. Ghose, and T. Hoshino, Leukemia in Hiroshima AtomicBomb Survivors, filood (in press); ibid., Technical ReportO2-59, Atomic Bomb Casualty Commission, March 15, 1959.



HEMPELMANNâ€”IonizingRadiation and Leukemia

the incidence increases as an inverse function of thedistance of the population groups from the hypo-center is well established. The study of Maloneyand Kastenbaum (16) published in 1955 indicatedthat the ratio of observed to expected cases even inthe lowest exposure zone (2,000-2,999 meters fromthe hypocenter) at Hiroshima was greater thanthat expected (1.5:1). In the most recent reportfrom Hiroshima, Heyssel and colleagues2 discussfully the meaning of the values for leukemia incidence in terms of the bias introduced by attemptsto estimate retrospectively the number of personsin each exposure zone. In their Table II, they showthe leukemia incidence in each zone according tothe number of exposed persons in each zone estimated on the basis of three census reports and ofone other selected group of survivors. Only in thecase of the so-called "reserve" group of the 1950

census, a small and highly biased group (whoseofficial and actual residences were not identical)does the leukemia incidence in the persons between 2,000 and 2,900 meters exceed that in twounexposed groups. These control groups are composed of 24,091 persons with a total of 168,119years-at-risk.

Heyssel and his colleagues also present datashowing the leukemia frequency according to additive gamma rays and neutron dose in the open airat various distances from the hypocenter. In thesecalculations, the latest dosage data obtained fromrecent tests were used, and a relative biological efficiency (RBE) of one was used for the neutrons.The authors estimate that 60 per cent of the peoplewere indoors at the time of the explosion and statethat this would reduce the air dose by 30-70 percent. They consider smaller dose increments(about a factor of two) per group than those usedformerly. Using only leukemia cases diagnosed between the years 1950 and 1958, they observed alinear relationship between incidence and the calculated open air dose of 177 rads or above. Thepoint representing the leukemia incidence of 3,605persons receiving a mean dose of 77 rads falls almost exactly on the line drawn through the pointsat higher dosage. Although twice the spontaneousincidence, the increase at this dosage is not significant. No cases of leukemia were observed in 3,512and in 1,305 people receiving an average dose of 34and 19 rads, respectively. They state that there isinsufficient data to permit a definitive analysis ofthe dose-response curve in the region below 77rads, a conclusion that seems well justified.

In addition to the studies on leukemia incidenceas a function of distance and dose, Heyssel and hiscolleagues present evidence showing that the latent period between exposure and development of

the disease is a function of the size of the dose.Their data also indicate that radiation exposurecauses an excess number of cases rather than anacceleration of the time of appearance of caseswhich would occur spontaneously. They also report that practically all cases in either the exposedor nonexposed persons were acute in nature or, ifchronic, then of the myelocytic type; chroniclymphocytic leukemia was rare in either group.

The recent report of the Japanese survivors ofthe atomic explosions appears to provide the mostreliable data to date on the subject, even thoughthe leukemia cases occurring before 1950 are excluded. Although errors have undoubtedly beenintroduced by establishing the so-called "closed"

population so long after the event, it is probablyimpossible to improve upon the samples. By rigorous definition of the sample in each exposurezone, the number of cases of leukemia is reduced,particularly in the zones more than 1,500 metersfrom the hypocenter. This makes the estimates ofleukemia incidence at these distances subject toconsiderable error. Future data on leukemia inHiroshima and perhaps information from Nagasaki now being compiled and analyzed may helpto clarify the question of the leukemogenic actionof radiation exposure at these distances.

The dose estimates, based on recent tests simulating the nuclear detonations during the war,would seem to have more basis in fact than anyformerly made. The authors point out, however,that there is still uncertainty in the estimates dueto errors in the precise location and shielding ofeach individual. For a number of other reasons, itis evident that there must be a considerable margin of error associated with the individual dosevalues. It has been pointed out by Bond3 that atleast 200 people survived in the exposure zonewhere the mean open air dose was calculated to be2,620 rads. Even allowing for shielding and assuming that these persons were located at theperiphery of the exposure zone, the doses receivedby these survivors according to these calculationsmust have been well in excess of that generallyconsidered to be 100 per cent lethal. In the lowdose region, a question may also be raised aboutthe exactness of the calculated doses. Many persons in the zone 2,000-2,499 meters where theaverage estimated dose was 6 rads reported a history of severe radiation symptoms (epilation, oro-pharyngeal lesions, and purpura) (15). This doseis far below that associated with radiation sicknessafter total-body exposure (18). Since persons evenfarther away were said to have similar symptoms,it is the consensus of a number of observers3 pres-

3V. C. Bond, personal communication.




ent at the time of the interviews with the patientsthat these symptoms are probably due to factorsother than radiation. Finally, it has been suggestedby Mole (21) that the use of the dose at the midpoint of the exposure zones as the mean dose of thepeople in that zone is misleading for reasons ofgeometry.

Since a number of assumptions is involved inthese calculations, e.g., the RBE for neutrons, theapplicability of data from the simulated tests tothe exposure conditions at Hiroshima, etc., it isperhaps best to accept even these latest dose estimates with reservation. It is obvious that the leukemia incidence regresses with diminishing dose,but how far down the dosage scale this regressionholds is not certain. Taking shielding into account,Heyssel and his colleagues4 believe that 50-100rads is the probable actual dose received by persons in the outermost zones showing an increasedleukemia incidence. They admit, however, thatthere may be a considerable error in their estimates.

RADIOLOGISTSRECEIVINGPARTIAL-BODYRADIATION(X-RAYS) OVERMANYYEARS

Since 1944, a number of studies have shown theincreased frequency of deaths from leukemia inAmerican radiologists (10, 17, 27). Most of thesereports are based on analyses of the obituaries appearing in the Journal of The American MedicalAssociation, a method of epidemiological studyopen to criticism. Nevertheless, it is clear fromthese studies that the incidence of leukemia deathsin American radiologists exceeds by far that expected in the general male population or that inother medical specialties. In contrast, excess casesof leukemia have not occurred in British radiologists who began to practice after 1921, and onlytwo cases occurred in radiologists in practice before this time (6). The absolute number of published cases of persons developing leukemia amongthe many thousands of x-ray workers exposed toradiation since the turn of the century is not large,only 87 well documented cases having been reported in the literature (15).

Although the fact that repeated irradiation withsmall doses of x-rays over a period of many yearscan be associated with a high leukemia incidenceis of great interest, it is almost impossible to makeany sensible statements about the relationship ofdose and leukemia frequency. Braestrup (1) estimates that a radiologist working with old-typex-ray equipment and few protective measures wasexposed to as much as 100 roentgens of x-rays peryear. He believes that exposure of individuals be-

4 Heyssel, personal communication.

fore 1930 was considerably higher than this,whereas, at the present time, it averages considerably less than 5 r per year. His estimate of thecumulative total exposure of a radiologist usingold-type x-ray units was of the order of 2,000 rduring 40 years of practice. In the study to bementioned later, Lewis (12) estimated the averageexposure of all radiologists to be 30 rad per yearor 1,200 r in 40 years.

Although it is apparent that a relatively fewold-time radiologists received large doses of radiation over a period of many years, the estimates ofthe average dose for all radiologists obviously aresubject to great error. In any given period, the exposure varied with the practice of each individual,his concern for safety, the type of equipment used,and his work load. How the gradual improvementin safety standards and the rapid increase in thenumber of radiologists influenced the average individual exposures can only be a matter of speculation. It is clear, however, that the scattered radiation from the low-voltage diagnostic equipmentresponsible for most of the exposure was poorlypenetrating in nature. The energy of this soft radiation was dissipated mainly in the tissues of thearms and hands of the radiologists. The abdomenand chest or head and upper chest received thenext highest dose depending upon protectivemeasures used. The exposures were intermittent,and the dose rate was low compared with those inthe previous three studies.

Aside from showing that repeated intermittentpartial-body exposure over many years is associated with an increased incidence of leukemia,study of American radiologists adds little to ourknowledge of the quantitative aspects of the relationship between x-ray exposure and the disease.The difference in leukemia frequency in Britishand American radiologists could be due to the highBritish standards of radiation protection introduced in the early days of radiology (6).

PRENATALEXPOSURETO DIAGNOSTICDOSES OFX-RAYSIN CHILDRENWITHLEUKEMIA

ANDCANCERRecently, Stewart and her colleagues (32) pub

lished an extensive survey of the retrospective interview type of 1,399 children dying of leukemiaor cancer. These children died before the age of10 in England and Wales during a 3-year period,1954â€”1956.One of their significant findings withrespect to radiation exposure was the frequency ofdiagnostic x-ray examinations (13.7 per cent) ofthe mothers of children with malignant disease.This was almost double the per cent (7.2) ofsuch examinations in the case of mothers of the



HEMPELMANNâ€”Ionizing Radiation and Leukemia 25

control children. The control children in this studywere matched with the study child for age, sex,and locality but were otherwise randomly chosenfrom the official registers of births. The averagenumber of films taken was 2.37 per child and theexposure has been calculated to be of the order of86-600 milliroentgens per film measured at thegonads of the fetus (33). If one assumes that radiation was the etiologic factor in the 6-7 per centexcess of the leukemia children receiving prenatalexposure, as the authors proposed, then sixteen toeighteen cases of leukemia per year would haveresulted from this diagnostic practice.

Four other retrospective studies in different sections of this country have attempted to obtain thesame information from the mothers of childrendying of leukemia and cancer. That of Ford andher colleagues in New Orleans (9) is concernedwith 78 leukemic children and 74 children withother malignant disease compared with 306 deadcontrols matched for color, age, and place of death.The findings of this study confirm the observationsmentioned above. Thus, 26.9 and 28.4 per cent ofthe children with leukemia and with other forms ofcancer, respectively, had been irradiated in utero,whereas only 18.3 per cent of the control childrenhad been so exposed.

The three following reports using other methodsfor selecting controls do not show the same excessof fetal irradiation in leukemic children. Polhemusand Koch (25) found no difference in the historyof prenatal irradiation in 251 children diagnosedas having leukemia in the Children's Hospital of

Los Angeles and in the same number of matchedcontrol children with nonorthopedic diseases onthe surgical service of the same hospital. In a current study of childhood leukemia in California,Kaplan and Moses (11) found that the number ofchildren with leukemia having a history of prenatal irradiation exceeded that of the untreatedsiblings. However, such an excess was not observed when the leukemic children were comparedwith healthy playmates. Murray and colleagues(22) found no difference in the history of prenatalexposure of 65 children with leukemia, 65 matcheddead controls, and the 175 living siblings of bothgroups.

In retrospective studies of this kind, the choiceof the control group may be crucial in reachingdefinitive conclusions. Questions can always beraised about the strict comparability of controlseven when selected with great care and forethought. It would seem that the studies as presented do not differentiate clearly between the association of leukemia and (a) the effect of themedical condition which justified the diagnostic

examination or (6) the effect of x-rays. Clearly,more studies, preferably prospective in type withmore than one control group, are needed to determine whether small doses of x-rays of the magnitude used in radiology are leukemogenic to thehuman fetus.

INDIVIDUALCASEREPORTSOFLEUKEMIAIN PERSONSEXPOSEDTORADIATION

There are a number of individual case reportsof leukemia occurring in persons who have beenexposed to the radiations from internally deposited radioactive materials. For example, four casesof leukemia (13) have been reported in patientsgiven thorotrast parenterally. Similarly, six casesof leukemia have been recorded in persons giventherapeutic doses of I131for hyperthyroidism (34).Since the total number of persons so treated is notknown, one cannot be certain that this representsmore than the expected number. In a recent report (34), it has been estimated that this is not anexcessive number of cases assuming that all havebeen reported. It is interesting to note, however,that five of the six cases occurred within 2 yearsafter treatment with the radioactive iodine. Thisis a much shorter latent period than that usuallyobserved when leukemia develops in persons exposed to ionizing radiation.

In a report on the frequency of diagnostic radio-graphic procedures in 828 cases of leukemia, Faber(8) states that adults with chronic myelogenousand acute leukemia have a history of having morefrequent diagnostic x-ray procedures than personswith chronic lymphocytic leukemia. He believesthat the exposure of the body to diagnostic dosesof x-rays must be partly responsible for causingthese forms of leukemia.

THE LEWIS' STUDY OF FOUR IRRADI

ATED POPULATIONSIn a stimulating and highly provocative article

published in Science, Lewis (12) compared theavailable data on leukemia incidence and radiationdoses in four of the irradiated groups mentionedabove. He attempted to assess the probability ofdeveloping leukemia on the basis of the calculateddose of radiation absorbed in the blood-formingtissues of each individual of the four groups. Using a number of assumptions concerning dosagesand correcting for the amount of tissue exposed,Lewis found that his "best" estimate of the probability of developing leukemia/individual/roent-gen/year is essentially the same in all groups,namely, 1-2 X 10~6. Even for the different exposure groups of the patients with ankylosing spon-dylitis and of the survivors of the atomic bomb-




ings, his calculation of the probability of developing leukemia per roentgen was approximately thesame. Because each roentgen appears to be equallyeffective in inducing leukemia, regardless of race,sex, age, and period of exposure, Lewis concludedthat the cases of leukemia in irradiated individualscould be regarded as the somatic counterpart to aradiation-induced gene mutation in germ plasm.

Reconsidering his calculations in the light ofmore recent information, the values for the probability for the different groups do not seem quite asclose as Lewis found. Using the new estimates ofleukemia frequency in Hiroshima, the probabilityvaries from 0.3 to 1.2 X 10~6in the different expo

sure zones (33). The negative probability in the2,000-2,999-meter zone does not seem meaningfulin view of the small size of the group and the absence of cases. Lewis' probability value calculated

for children with thymic enlargement and forAmerican radiologists should be modified, becausehis estimates of the average dose to the entireblood-forming tissues are incorrect. Since the average dose administered to each child was 250 roentgens and since less than a fourth or fifth of thebody was exposed, the probability in their casesshould be at least 4-5 X 10~6rather than 1 X 10~6.According to Lewis' estimates of the average ex

posure for the radiologists, but with the assumption that the soft x-rays irradiated only one-halfto one-third of the blood-forming tissues, the probability becomes at least 4-6 X 10~6.In the case of

the children irradiated prenatally, which Lewis didnot consider, the probability turns out to be of theorder of 10-11 X 10~6.

The minor corrections of Lewis' calculation do

not alter the fact that dividing the leukemia frequency by the estimated dose results in a remarkably constant number. Whether it is justifiable toinclude the radiologists in this study is debatablesince the dose estimates represent nothing morereliable than a guess. The recent finding of Russell(26) that the frequency of genetic mutations varieswith the dose rate raises the question whether ornot a constant probability of producing leukemiaper roentgen absorbed in the blood-forming tissueshould be considered as evidence for a somaticmutation being responsible for the development ofthe disease. This may be unimportant in the present studies because the dose rate is relatively highin all cases. The lowest dose rate presumably occurred in the radiologists who were exposed toscattered x-ray mainly during fluoroscopic examinations. Even in these cases, the exposure rate wasprobably high compared with background radiation. A final point to be considered is that the number of persons receiving less than 500 roentgens is

relatively small in the case of the spondylitic seriesand the radiologists, and, in the latter series, thedose is highly uncertain. Therefore, it would seemthat Lewis's probability values do not apply to

persons exposed to less than 500 roentgens exceptfor children and for some of the Japanese.

ADDENDUMSince this article was submitted for publication, the results

of two other studies of a similar nature have been brought tothe attention of the author. Latourette and Hodes (Am. J.Roentgenol., 82:667, 1959) studied 867 out of 958 patientstreated for thymic enlargement at the University Hospital inAnn Arbor, Michigan, mostly before 1942. Seven hundred andeighty-three received less than 200 roentgens, usually througha 10 X 10 cm. port. One of the children of the families contacted had died of leukemia and another of lymphosarcoma. Inanother, as yet unpublished, study, Saenger, Silverman, Sterling and Turner (personal communication) interviewed thefamilies of 1,644 of 2,280 children treated with x-rays for various benign conditions in Cincinnati, Ohio. One of the tracedgroup had died of leukemia, ten had cancer of the thyroid, fivehad other types of malignant disease, and a number of othershad benign neoplasia. Forty per cent of the children had received less than 200 roentgens, 62 per cent less than 400 roentgens, and 20 per cent had an unknown dose. In 31 per cent ofthe cases, the port size was less than 75 sq. cm., and in almostall others with known exposure factors, the port was less than100 sq. cm. Approximately one-third of the patients were treated for thymic enlargement, and another third received treatment to the head and neck.

REFERENCES1. BRAESTRUP,C. B. Past and Present Exposure of Radiolo

gists from the Point of View of Life Expectancy. Am. J.Roentgenol., 78:988, 1957.

2. BRUES,A. M. Critique of The Linear Theory of Carcino-genesis. Science, 128:693, 1958.

3. CONTI,E. A., and PATTON,G. B. Study of The Thymus in7,400 Consecutive Newborn Infants. Am. J. Obst. & Gyn.,66:885, 1948.

4. CONTI,E. A.; PATTON,G. B.; CONTI,J. E.; and HEMPEL-MANN,L. H. The Present Health of Children Given X-RayTreatments to Their Anterior Mediastinum in Infancy.Radiology (in press).

5. COUKT-BROWN,W. M., and DOLL, R. Leukemia andAplastic Anemias in Aukylosing Spondylitis. Med. Res.Council Report 295. London: Her Majesty's Stationery

Office, 1957.6. . Expectation of Life and Mortality from Cancer

among British Radiologists. Brit. M. J., 2:181, 1958.7. . Adult Leukemia. Ibid., 1:1063, 1959.8. FABER,M. Radiation-induced Leukemia in Denmark. In:

Adv. in Radiobiology. Springfield: Charles C Thomas,1957.

9. FORD,D. D.; PATERSON,J. C. S.; and THUNTTNG,W. L.Fetal Exposure to Diagnostic X-rays and Leukemia andOther Malignant Disease in Childhood. J. Nat. CancerInst.,22:1093, 1959.

10. HENSHAW,P. S., and HAWKINS,J. W. Leukemia Incidencein Physicians. J. Nat. Cancer Inst., 4:339, 1944.

11. KAPLAN,H. S. The Evaluation of The Somatic and GeneticHazards of the Medical Uses of Radiation. Am. J. Roentgenol., 80:696, 1958.

12. LEWIS, E. B. Leukemia and Ionizing Radiation. Science,126:965, 1957.



HEMPELMANNâ€”IonizingRadiation and Leukemia 27

18. LOONET,W. B., and COLODZIN,M. Late Follow-up Studiesafter Internal Deposition of Radioactive Materials.J.A.M.A., 160:1, 1956.

14. LOUTIT,J. F. The Effects of Radiation on The Individual.Practitioner, 181:553, 1958.

15. MALONEY,W. C. Induction of Leukemia in Man by Radiation. Radiation Biology and Cancer. Austin: Univ. ofTexas Press, 1959.

16. MALONEY,W. C., and KASTENBAUM,M. A. LeukemogenicEffects of Ionizing Radiation in Atomic Bomb Survivors inHiroshima City. Science, 121:808, 1955.

17. MARCH, H. C. Leukemia in Radiologists in a 20-yearPeriod. Am. J. M. Se., 220:282, 1950.

18. MILLER,L. S.; FLETCHER,G. H.; and GERSTNER,H. B.Radiobiologie Observations on Cancer Patients Treatedwith Whole Body Irradiation. Radiation Research, 8:150,1958.

19. MOLE, R. H. Radiation and Leukemia. Lancet, 2:192,1957.

20. . The Dose Response Relationship in RadiationCarcinogenesis. Brit. M. Bull., 14:184, 1958.

21. . Some Aspects of Mammalian Radiobiology. Radiation Research, Suppl. 1, 1959.

22. MURRAY,R.; HECKEL,P.; and HEMPELMANN,L. H. Leukemia in Children Exposed to Ionizing Radiation. NewEngland J. Med. (in press).

23. O'CONNELL,D. Heredity in Ankylosing Spondylitis. Arch.

Int. Med., 60:1115, 1959.

24. POCHIN,E. E. Radiation and Leukemia. Lancet, 1:51,1959.

25. POLHEMUS,D. W., and KOCH,R. Leukemia and MedicalRadiation. Pediatrics, April, 1959.

26. RUSSELL,W. L.; RUSSELL,L. B.; and KELLY,E. M. Radiation Dose Rate and Mutation Frequency. Science, 128:1546, 1958.

27. SCHWARTZ,E. E., and UPTON,A. C. Factors Influencingthe Incidence of Leukemia; Special Considerations of theRole of Ionizing Radiation. Blood, 13:845, 1958.

28. SIMPSON,C. L. Radiation-induced Neoplasms in Man.Radiation Biology and Cancer, p. 336. Austin: Univ. ofTexas Press, 1959.

29. SIMPSON,C. L., and HEMPELMANN,L. H. The Associationof Tumors and Roentgen-Ray Treatment of the Thorax inInfancy. Cancer, 10:42, 1957.

30. SIMPSON,C. L.; HEMPELMANN,L. H.; and FULLER,L. M.Neoplasia in Children Treated with X-rays in Infancy forThymic Enlargement. Radiology, 64:840, 1955.

31. SNEGIREFF,L. S. The Elusiveness of Neoplasia FollowingRoentgen Therapy in Childhood. Radiology, 72:508, 1959.

32 STEWART,A.; WEBB, J.; and HEWITT,J. A Survey ofChildhood Malignancies. Brit. M. J., 1:1495, 1958.

33. United Nations Scientific Committee Report on TheEffects of Atomic Radiation. Suppl. 17 (A/3838), 1958.

34. VELLER,V., and HOFER, R. Acute Leukemia FollowingRadioiodine Therapy of Thyrotoxicosis. Brit. J. Radiology, 32:263, 1959.



1960;20:18-27. Cancer Res L. H. Hempelmann Ionizing RadiationEpidemiological Studies of Leukemia in Persons Exposed to

Updated version

http://cancerres.aacrjournals.org/content/20/1/18

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/20/1/18To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



epidemiological studies of leukemia in persons …...epidemiological studies of leukemia in persons...

Documents